HIV-1 is a global pandemic with about 39 million people infected. In India, 2.9 million people are infected and about 2 lakh new infections have been reported last year. To date, there is no cure for HIV/AIDS. Current treatment, which is associated with serious side effects, only delays the onset of AIDS and death. Thus, HIV/AIDS is responsible for a global health concern imposing significant healthcare costs, especially in low- and middle-income regions such as India and Africa, and a marked loss of quality of life to infected individuals. Understanding factors impacting vaccine design and drug development via mathematical modelling of HIV-1 transmission, evolution and pathogenesis and discerning the subtype and region specific differences are a crucial part of the overall strategy of reducing the burden of HIV/AIDS. The strain dominant in India is HIV-1 subtype C (HIV-1C). Treatment guidelines have largely been based on studies on HIV-1 subtype B (HIV-1B), dominant in the west. In this thesis, we have attempted to understand the dynamics of the spread of HIV-1C, leading to new guidelines and intervention strategies applicable to India. We have for the first time estimated the basic reproductive ratio, R0, of HIV-1 subtype C (HIV-1C), a proxy for its fitness and virulence, using clinical data of infected patients from India. We employed measurements of viral load decay dynamics during treatment and estimated R0, and the critical efficacy, εc, for successful treatment of HIV-1C infection. Clinical data showed that the viral load in patients in India was significantly higher than in the west. Yet, in 6 months following the start of treatment, 87.5% had undetectable viral load, indicating an excellent response to ART, comparable to the west. We analyzed the clinical data using a mathematical model and estimated the median R0 to be 5.3. The corresponding εc was ∼0.8. These estimates of R0 and εc are smaller than current estimates for HIV-1B, suggesting that HIV-1C exhibits lower in vivo fitness compared to HIV-1B, which allows successful treatment despite high baseline viral loads. New treatment guidelines thus emerge that are less stringent than in the west. HIV-1C is far more prevalent globally than HIV-1B. This is surprising in light of our findings above of a lower fitness of HIV-1C than HIV-1B. To understand this observation, we next developed a mechanistic paradigm of HIV-1 transmission. HIV-1 has been hypothesized to optimize its transmission potential (TP) in an infected population by modulating its steady state viral load (VSS), a robust marker of virulence. The mechanism of this optimization is paradoxical and poorly understood given that HIV-1 mutates rapidly in vivo in response to selection pressure by the host immune system. We hypothesize that the HIV-1 TP is not solely a function of VSS as proposed earlier, but a function of two variables - VSS and R0, which function such that R0 is optimized within an infected individual in response to the immune system while VSS is optimized across individuals such that transmission is optimized. On this TP(VSS, R0) landscape, we find that HIV-1C lies closer to the optimum than HIV-1B, suggesting an explanation for the global spread of HIV-1C. This leads to the intriguing implication that the lower virulence of HIV-1C may be because it has evolved more along the TP(VSS, R0) landscape than HIV-1B. Lastly, we examined the role of recombination on HIV-1 adaptation. Following transmission, HIV-1 adapts in the new host by acquiring mutations that allow it to escape from the host immune response at multiple epitopes. It also reverts mutations associated with epitopes targeted in the transmitting host but not in the new host. Moreover, escape mutations are often associated with additional compensatory mutations that partially recover fitness costs. It is unclear whether recombination expedites this process of multi-locus adaptation. To elucidate the role of recombination, we constructed a detailed population dynamics model that integrates viral dynamics, host immune response at multiple epitopes through cytotoxic T lymphocytes, and viral evolution driven by mutation, recombination, and selection. Using this model, we computed the expected waiting time until the emergence of the strain that has gained escape and compensatory mutations against the new host’s immune response, and reverted these mutations at epitopes no longer targeted. We found that depending on the underlying fitness landscape, shaped by both costs and benefits of mutations, adaptation proceeds via distinct dominant pathways with different effects of recombination, in particular distinguishing escape and reversion. Specifically, recombination tends to delay adaptation when a purely uphill fitness landscape is accessible at each epitope, and accelerate it when a fitness valley is associated with each epitope. Our study points to the importance of recombination in shaping the adaptation of HIV-1 following its transmission to new hosts, a process central to T cell-based vaccine strategies.